Remove ordereddict io.ascii - #16240

Thank you for your contribution to Astropy! 🌌 This checklist is meant to remind the package maintainers who will review this pull request of some common things to look for.

• Do the proposed changes actually accomplish desired goals?
• Do the proposed changes follow the Astropy coding guidelines?
• Are tests added/updated as required? If so, do they follow the Astropy testing guidelines?
• Are docs added/updated as required? If so, do they follow the Astropy documentation guidelines?
• Is rebase and/or squash necessary? If so, please provide the author with appropriate instructions. Also see instructions for rebase and squash.
• Did the CI pass? If no, are the failures related? If you need to run daily and weekly cron jobs as part of the PR, please apply the "Extra CI" label. Codestyle issues can be fixed by the bot.
• Is a change log needed? If yes, did the change log check pass? If no, add the "no-changelog-entry-needed" label. If this is a manual backport, use the "skip-changelog-checks" label unless special changelog handling is necessary.
• Is this a big PR that makes a "What's new?" entry worthwhile and if so, is (1) a "what's new" entry included in this PR and (2) the "whatsnew-needed" label applied?
• At the time of adding the milestone, if the milestone set requires a backport to release branch(es), apply the appropriate "backport-X.Y.x" label(s) before merge.

If it's building on top of a draft PR I suggest drafting this one too !

@MridulS - the change to MetaData is likely to cause too many problems. The YAML representation of this is bigger than just astropy. A few things to note:

• In YAML the key order is explicitly not ordered. So even if one Python implementation happens to write them out in order this does not guarantee anything with other readers. https://yaml.org/spec/1.2.2/#3221-mapping-key-order
• The ECSV spec explicitly calls out !!omap as the serialization type for meta. This cannot be changed without a change to APE-6. There are other non-Python reader/writers for ECSV so such a change is not likely to be approved unless there is a very strong driver. https://github.com/astropy/astropy-APEs/blob/main/APE6.rst

The ECSV spec explicitly calls out !!omap as the serialization type for meta. This cannot be changed without a change to APE-6. There are other non-Python reader/writers for ECSV so such a change is not likely to be approved unless there is a very strong driver. https://github.com/astropy/astropy-APEs/blob/main/APE6.rst

Woops, did not see that! I guess OrderedDict needs to be there. I don't think it is going away from CPython anytime soon. The only main gain here would be the performance gain. Vanilla dict is supposed to be much more efficient than OrderedDict

The only main gain here would be the performance gain. Vanilla dict is supposed to be much more efficient than OrderedDict

I haven't done profiling but I would be a bit surprised if meta handling is a substantial factor in performance for many real-world applications.

I haven't done profiling but I would be a bit surprised if meta handling is a substantial factor in performance for many real-world applications.

Yeah I was thinking more about Table wide usage of OrderedDict, like in TableColumns too, but maybe the abstraction is far enough. I should run some benchmarks :)

@MridulS - I was poking around the code and wondered about an idea to use OrderedDict only at the interface with YAML, in that way preserving the !!omap tag. I have not tested this, but you could try these changes:

➜  astropy git:(main) ✗ git diff
diff --git a/astropy/table/meta.py b/astropy/table/meta.py
index 3af1c64cab..20883641c5 100644
--- a/astropy/table/meta.py
+++ b/astropy/table/meta.py
@@ -76,7 +76,7 @@ def _construct_odict(load, node):
       >>> yaml.load('''!!omap [ foo: bar, mumble: quux, baz : gorp ]''')  # doctest: +SKIP
       OrderedDict([('foo', 'bar'), ('mumble', 'quux'), ('baz', 'gorp')])
     """
-    omap = OrderedDict()
+    omap = {}
     yield omap
     if not isinstance(node, yaml.SequenceNode):
         raise yaml.constructor.ConstructorError(
@@ -269,7 +269,7 @@ def _get_col_attributes(col):
         ("unit", lambda x: x is not None, str),
         ("format", lambda x: x is not None, None),
         ("description", lambda x: x is not None, None),
-        ("meta", lambda x: x, None),
+        ("meta", lambda x: OrderedDict(x), None),
     ):
         col_attr = getattr(col.info, attr)
         if nontrivial(col_attr):
@@ -302,7 +302,7 @@ def get_yaml_from_table(table):
     """
     header = {"cols": list(table.columns.values())}
     if table.meta:
-        header["meta"] = table.meta
+        header["meta"] = OrderedDict(table.meta)
 
     return get_yaml_from_header(header)

With the changes above, you should also revert the changes that add sort_keys=False. That should no longer be necessary and this will hopefully reduce the change footprint even further. The goal is for this PR to be minimally impacting and not break any unit or doc tests. (That's the goal, we'll see about reality).